Restorative
Policing
Experiment
The Bethlehem Pennsylvania Police
Family Group Conferencing Project
Paul McCold, Ph.D.
&
Benjamin Wachtel
Community Service Foundation
Pipersville, PA
May 1998
This project was supported under award number 95-IJ-CX-0042 from the National Insti-tute of Justice, Office of Justice Programs, U.S. Department of Justice. Points of view in this document are those of the authors and do not necessarily represent the official position of the U.S. Department of Justice.
The authors gratefully acknowledge the following individuals for their contributions to this project: Captain John Stahr, Project Liaison, Bethlehem Police Department, without whom this study would not have been possible. John provided the operational and motivational support responsible for Operation P.R.O.J.E.C.T.
Ted Wachtel, Executive Director, Community Service Foundation, for his collaboration in launching the Restorative Policing Experiment and for providing the essentials of office space and support.
Police Commissioner Eugene Learn, former Police Commissioner John W. Yerk and the Bethlehem Police Department, for their assistance and ongoing support and cooperation. The Restorative Policing Experiment research advisory board, for their valuable input and
guidance:
• Howard Zehr, Director, Office of Criminal Justice, Mennonite Central Committee • James Anderson, Executive Director, Pennsylvania Juvenile Court Judges Commission • Ruth Williams, Juvenile Justice Program Manager, Pennsylvania Commission on Crime
and Delinquency
• Ronald Sharp, Director of Psychological Services, Alternative Rehabilitation Communities • Henry Sontheimer, Senior Evaluation Analyst, Pennsylvania Commission on Crime and
Delinquency
• Philip Harris, Professor of Criminal Justice, Temple University
• Mary Achilles, Pennsylvania Victim Advocate and Director of Victim Services, Pennsylvania Department of Corrections
Sergeant Terry O’Connell, New South Wales Police Service, master facilitator and the father of restorative policing, whose standards for conferencing formed the basis of the program protocol used in this research.
John Braithwaite and Heather Strang, Australia National University, Lawrence Sherman, University of Maryland, and the enthusiastic RISE researchers and advisors for their generous sharing of research instruments and for providing invaluable insights and support into our project’s implementation and analyses.
The REAL JUSTICE team for their daily support and technical expertise and for their ongoing success in putting what they preach into practice.
Mark Umbreit, Center for Restorative Justice & Mediation, for his pioneering work evaluating mediation, which set a high standard for restorative justice research. Mark’s work made this study’s comparisons of conferencing with VOM possible.
Captain Phillip Richardson, Bethlehem Police Department; Nick Melnick, Electronic Data Interchange Coordinator, Administrative Office of Pennsylvania Courts; and the staff at Northampton County and Lehigh County Juvenile Probation for their assistance in data collection.
Criminal justice officials from Lehigh and Northampton counties for their support and cooperation:
• Judge William Moran, Northampton County • Judge Edward Reibman, Lehigh County
• John Morganelli, Northampton County District Attorney • Robert Steinberg, former Lehigh County District Attorney
• Dennis Galligani, Chief, Northampton County Juvenile Probation • Paul Werrell, Chief, Lehigh County Juvenile Probation
• Nancy Matos-Gonzalez, District Justice, Bethlehem • Elizabeth Romig, District Justice, Bethlehem
• Barbara Schlegel, District Justice, Bethlehem • James Stocklas, District Justice, Bethlehem • Thomas Murphy, District Justice, Bethlehem
This is a report on the Bethlehem Pennsylvania Police Family Group Conferencing Project. First-time moderately serious juvenile offenders were randomly assigned either to formal adjudication or to a diversionary “restorative policing” process called family group conferenc-ing. Police-based family group conferencing employs trained police officers to facilitate a meet-ing attended by juvenile offenders, their victims, and their respective family and friends, to discuss the harm caused by the offender’s actions and to develop an agreement to repair the harm. Victim and offender participation is voluntary. The effect of the program was mea-sured through surveys of victims, offenders, offender’s parents and police officers and by examining outcomes of conferences and formal adjudication. Results are related to six ques-tions about restorative policing. Findings include: 42% participation rate, 100% of confer-ences (n=67) reaching an agreement, 94% of offenders (n=80) fully complying with agree-ments, and participant satisfaction and sense of fairness exceeding 96%. Results suggests that recidivism was more a function of offenders choice to participate than the effects of the conferencing, per se. Violent offenders participating in conferences had lower rearrest rates than violent offenders declining to participate, but this was not true for property offenders.
Table of Contents
EXECUTIVE SUMMARY ... 1
1. RESTORATIVE POLICING ... 7
2. THE BETHLEHEM EXPERIMENT ... 15
3. CONFERENCE OBSERVATIONS ... 27
4. POLICE SURVEYS ... 39
5. PARTICIPANT SURVEYS ... 47
V
ICTIM SURVEY RESULTS... 51
O
FFENDERSURVEY RESULTS... 58
P
ARENT SURVEYRESULTS... 64
6. RECIDIVISM ... 73
7. SYSTEMIC RESPONSES ... 79
8. COMPARATIVE ANALYSES ... 89
9. CONCLUSIONS ... 103
L
IMITATIONS OF CURRENT RESEARCH... 110
APPENDICES ... 115
List of Exhibits
Exhibits page #
Summary
Exhibit S1: Satisfaction with handling of case ... 5
Exhibit S2: Participation rate comparison with VOM ... 6
Technical Report The Bethlehem Experiment Exhibit 1: Proportion of total juvenile arrests disqualified from study ... 18
Exhibit 2: Offender eligibility categories ... 18
Exhibit 3: Random assignment results - Cases ... 19
Exhibit 4: Random assignment results - Offenders ... 19
Exhibit 5: Treatment group participation rates ... 19
Exhibit 6: Arrest charges for offenders included in study by experimental group ... 20
Exhibit 7: Arrest charge categories for offenders included in study by experimental group ... 20
Exhibit 8: Experimental group comparisons ... 21
Exhibit 9a: Reasons for cases declining to participate ... 22
Exhibit 9b: Corrected Participation Rates ... 22
Exhibit 10: Participation rates of treatment-selected offenders ... 23
Exhibit 11: Offender case disposition by crime type and control group ... 24
Exhibit 12: Magistrate findings for cases in study disposed by magistrate court ... 24
Conference Observations Exhibit 13: Type of victims ... 29
Exhibit 14: Mean number of inappropriate coordinator responses by period of experiment ... 31
Exhibit 15: Mean score for distinguishing deed from doer by period of experiment ... 31
Exhibit 16: Mean facilitator grade by period of experiment ... 32
Exhibit 17: Mean coordinator grade by conference sequence ... 33
Exhibit 18: Mean number of participants by conference sequence ... 33
Exhibit 19: Observer ratings of most punitive participant ... 35
Exhibit 20: Conference agreement terms ... 35
Exhibit 22: Reliability of Hassles and Uplifts Scales ... 41-42
Exhibit 23: Police attitude scales reliability ... 41
Exhibit 24: Police orientation scales reliability ... 42
Exhibit 25: Conferencing scales reliability ... 43
Exhibit 26: Police survey response rates ... 44
Exhibit 27: Knowledge of conferencing by experimental period ... 45
Exhibit 28: Mean change in orientation toward the use of force ... 45
Exhibit 29: Mean change in perception of community cooperation ... 46
Exhibit 30: Mean change in crime control orientation ... 46
Participant Surveys Exhibit 31: Case dispositions of offenders in study ... 47
Exhibit 32: Participant survey response rates ... 50
Exhibit 33: Victim satisfaction ... 52
Exhibit 34: Victims experiencing fairness ... 52
Exhibit 35: Victims agreeing offender was held accountable ... 52
Exhibit 36: Victims agreeing their opinion was considered ... 52
Exhibit 37: Court victims attitudes toward offense and offender ... 53
Exhibit 38: Conferenced victims perceptions ... 54
Exhibit 39: Conferenced victims agreeing with statements about conferencing ... 55
Exhibit 40: Importance of issues for victims ... 56
Exhibit 41: Offender satisfaction ... 59
Exhibit 42: Offender attitude toward victim ... 60
Exhibit 43: Court offenders attitudes toward victims ... 60
Exhibit 44: Conference offenders perceptions of conferencing ... 61
Exhibit 45: Conferenced offenders attitudes toward conferencing ... 62
Exhibit 46: Importance of issues for offenders ... 62
Exhibit 47: Offender’s parent satisfaction ... 65
Exhibit 48: Offender’s parents reporting sense of fairness ... 65
Exhibit 49: Offender’s parents agreeing their opinion was considered ... 66
Exhibit 50: Offender’s parents attitudes toward conferencing ... 67
Exhibit 51: Offender’s parents perceptions of conferencing ... 68
Exhibit 52: Importance of issues for offender’s parents ... 69
Offender Recidivism Exhibit 53: Rearrest rates by days of exposure by crime type ... 75-76 Exhibit 54a: Rearrest rates for violent offenders ... 76
Exhibit 54b: Rearrest rates for property offenders ... 76
Exhibit 55: Rearrest rates for offenders declining to participate by reasons for decline ... 77
Exhibit 56: Monthly juvenile arrests ... 80
Exhibit 57: Juvenile arrests 1995 to 10/1997 ... 80
Exhibit 58: Proportion of eligible cases selected during experimental period ... 81
Exhibit 59a: Rearrest rates - all eligible property offenders ... 82
Exhibit 59b: Rearrest rates - all eligible violent offenders ... 82
Exhibit 60: Rearrest rates - total juvenile arrests by crime ... 83
Exhibit 61a: Rearrest rates - property offenders by eligibility category ... 83
Exhibit 61b: Rearrest rates - violent offenders by eligibility category ... 84
Exhibit 62: Total juvenile arrests by experimental period by eligibility category ... 84
Exhibit 63: Non-selected eligible and selected juvenile arrests by month ... 85
Exhibit 64: Arrests handled informally and selected arrests by month ... 85
Exhibit 65: Disposition of eligible cases by experimental period ... 85
Exhibit 66: Disposition of cases not in study by experimental period ... 86
Exhibit 67: Percent payment ordered magistrate cases by experimental period ... 86
Exhibit 68: Mean payment ordered - magistrate cases by experimental period ... 86
Exhibit 69: Disposition comparison - magistrate cases by inclusion status and crime type ... 86
Exhibit 70: Percent payment ordered - magistrate cases by inclusion status and crime type ... 87
Exhibit 71: Mean payment ordered - magistrate cases by experimental period ... 87
Comparative Analysis Exhibit 72: Disposition of court-assigned cases in study ... 93
Exhibit 73: Percent of cases disposed via guilty plea ... 93
Exhibit 74: Proportion of offenders paying monetary costs by treatment group ... 94
Exhibit 75: Mean monetary costs for offenders paying costs ... 95
Exhibit 76: Participation rates comparison to VOM ... 95
Exhibit 77: Victim satisfaction comparison to VOM ... 96
Exhibit 78: Offender satisfaction comparison to VOM ... 96
Exhibit 79: Victim sense of fairness comparison to VOM ... 97
Exhibit 80: Offender sense of fairness comparison to VOM ... 97
Exhibit 81: Crime victim’s ratings of process comparison to VOM ... 99
Exhibit 82: Criminal offender’s ratings of process comparison to VOM ... 99
Bethlehem Pennsylvania Police Family Group Conferencing Project*
EXECUTIVE SUMMARY
Restorative justice is the latest trend in criminal justice practice that contains the seeds of a radically different paradigm on crime and justice than the traditional deter-rence or desert-based approaches. This report is an evaluation of one restorative justice program operated by the police in Bethlehem, Pennsylvania, a mid-sized American city whose justice practices are typical of thousands of such communities across the country. Although developed independently from the restorative justice movement, family group conferencing is considered an important new development in restorative justice practice as a means of dealing more effectively with young offenders by diverting them from court and involving their extended families and victims in addressing their wrong-doing. Originating in New Zealand in 1989, conferencing was substantially revised as a community policing technique in Wagga Wagga, New South Wales, Australia, in 1991. This was the first program to directly involve a justice official in conducting restorative justice, and has since broadened to include school officials, probation officers and others. The “Wagga model” was introduced to North America in 1995 by the Real Justice® orga-nization, and more than 2,000 police, probation officers, educators and others in the United States and Canada have now been trained as conference facilitators.
Purpose
The Bethlehem Pennsylvania Police Family Group Conferencing Project was de-signed to answer six programmatic questions about police-based conferencing as it is being applied in the United States.
1. Can typical American police officers conduct conferences consistent with due process and restorative justice principles?
2. Does involvement in conferencing transform police attitudes, organizational culture and role perceptions?
3. Does conferencing produce conflict-reducing outcomes by helping to solve ongoing problems and reduce recidivism?
4. Will victims, offenders and the community accept a police-based restorative justice response?
5. Does the introduction of diversionary conferencing alter the case processing of juvenile offenders (e.g., net-widening)?
6. How does police-based conferencing compare to the existing system and to other restorative justice practices?
*This project was supported under award number 95-IJ-CX-0042 from the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice. Points of view in this document are those of the authors and do not necessarily represent the official position of the U.S. Department of Justice.
Methods
In October 1995, 20 full-time police officers volunteered to be trained and conduct conferences. Over an 18-month period, first-time juvenile offenders arrested for selected misdemeanor and summary offenses were randomly assigned either to formal adjudica-tion or to a diversionary restorative policing conference. Cases were blocked by crime type: crimes primarily directed against the person (violent offenses) and crimes primarily directed against property (property offenses).
The effect of the program was measured through surveys of victims, offenders and offenders’ parents. Additional data was obtained from direct observations of conferences and review of official police and court records. Two department-wide surveys were con-ducted, prior to the first conference and again after 18 months of program operation. Officer attitudes on a wide range of questions about their work environment and the nature of policing were matched by officer for pre- and post-test comparisons.
During the course of the experiment, 215 criminal incidents involving the arrests of 292 juveniles qualified for the study, representing 23% of all juvenile offenders ar-rested in Bethlehem during the time period. These included 75 violent crimes and 140 property crimes. A store was the victim in 76% of the property cases and a school was the victim in 29% of the violent cases.
Participation in the program was voluntary, creating three groups of subjects: (1) statistical control group (n=68 property, 35 violent), (2) selected for conferencing and participating (n=56 property, 24 violent), and (3) selected for conferencing but not partici-pating (n=57 property, 52 violent). Conferences for violent offenses were conducted in 32% of cases selected for the treatment group, and in 50% of property cases, for an overall raw participation rate of 42% (proportion of conferences to cases selected). Offenders were much more likely to decline in property offenses and victims more likely in violent offenses. Among crime seriousness, number of charges, age, race, and gender of offender, only gender was significantly related to participation rate, and only among violent of-fenders with females participating at twice the rate as their male counterparts.
In spite of the probable self-selection bias in the treatment group, the generaliz-ability of the sample was maintained. However, for this experiment to demonstrate a recidivism reduction or an improvement over magistrate court cases, differences had to be strong enough to be measured across the entire treatment-selected group, even though less than half received the treatment. Questions about how police conducted conferences, whether this affected their culture, whether the community will accept the program, and how the program affected case processing do not require equivalent comparison groups and are unaffected by the threat to the internal validity of the experimental design.
Results
1. Can typical American police officers conduct conferences consistent with due process
and restorative justice principles?
There was an initial tendency among some officers to lecture the offender or influence the agreement in conferences. While they easily picked up the mechanics of the scripted process, an addi-tional in-service training was necessary early in the experiment to reinforce the reintegrative intention of conferences. Average grades for overall compliance with protocol improved significantly following the in-service training, from 80% to 89%.
In general, officers did a sufficient but not exemplary job in adhering to principles of restorative justice and ensuring due process. In spite of this, more than 96% of participants said they were satisfied with how their cases were handled and perceived the process as fair, more than 94% would choose to do the conference again, and more than 92% would recommend conferences to others. These results, which are consistent with the earlier evaluation of police conferencing in Australia, lend support to the generalizability of the Australian findings to police-based conferencing in the United States.
2. Does conferencing transform police attitudes, organizational culture and role
percep-tions?
There were no significant changes in overall police attitudes, organizational culture or role perceptions. Paired t-tests of pre- and post-test scores failed to detect any department-wide changes in attitudes during the experimental period. Thus, conferencing cannot be said to have had a significant impact on changing overall police attitudes toward their activities or the role of police.
The officers who had conducted conferences did show a significant increase in their percep-tions of community cooperation and a decrease in their orientation toward a crime control approach to policing. Thus the whole effect of conferencing was to cause a few officers who were positively dis-posed to community policing to become more supportive of such approaches.
3. Does conferencing produce conflict-reducing outcomes by helping to solve ongoing
problems and reduce recidivism?
Reducing offender recidivism is one measure of the capacity of restorative approaches to ad-dress the important needs created by a criminal offense. A reduction in re-offending is not the primary purpose, as in deterrence theory, but is one of a number of goals for the restorative response to crime. It is assumed that holding offenders accountable to their victims to repair the harm caused should increase offender empathy and thereby lead to a reduction in offending behavior.
Results indicate that lower recidivism for those participating in the program was more a func-tion of the offender’s choice to participate than the effects of the conference, per se. Violent offenders participating in conferences had significantly lower 12-month rearrest rates (20%) than those who declined to participate (48%). However, the control group rearrest rate (35%) was almost exactly be-tween the treatment-selected groups, indicating that there was little additional treatment effect beyond
a self-selection effect. Recidivism rates for the property offenders suggests that any self-selection ef-fect was transitory. There were significant differences for the decline and conference property offend-ers from 30 to150 days, however, these differences in the rearrest rates were not significant by 12 months.
The universal ability of conference participants to come up with mutually acceptable agree-ments demonstrates that conferences are useful in facilitating a collective, community-based solution to these criminal problems. The 94% offender compliance with the terms of the agreements supports the conclusion that these cases were resolved in a manner satisfactory to all participants.
It appears that any reductions in recidivism are the result of the voluntary program diverting from formal processing those juveniles who are least likely to re-offend in the first place. Presumably this is the goal of any good diversion program and, in this regard, the program was successful.
4. Will victims, offenders and the community accept a police-based restorative justice
response?
Victims participating in conferences said that they felt participating in the conference was their own choice (96%); they would recommend conferences to others (92%); they would choose a confer-ence if they had to do it over again (94%); meeting with the offender was helpful (93%); the tone of the conference was basically friendly (94%); the offender apologized (96%); and conferences should be offered to all victims (81%).
Offenders who participated in conferences said that it was their own choice to participate (92%); they would recommend conferencing to others (92%); if they had to do it over again, they would choose to participate (94%); meeting with the victim was helpful (100%); and the tone of the confer-ence was friendly (96%).
Nearly all parents of conferenced offenders said they would recommend conferencing to others (97%), would choose to participate in a conference if they had to do it over again (94%), thought that meeting the victim was helpful (97%), and that they had a positive or very positive attitude toward the conference (91%).
Victims, offenders and parents of offenders were consistently satisfied with the conferencing process and perceived the process and the outcomes as fair. Nearly all respondents indicated they would choose to participate in the program again and would recommend it to others facing similar trouble. While a majority of offenders declined to participate, a very high proportion of victims, of-fenders and ofof-fenders’ parents who did participate accepted this police-based restorative justice pro-cess.
5. How does the introduction of conferencing alter the case processing of juvenile
of-fenders?
There was no apparent change in overall arrest patterns for juvenile offenders during the ex-perimental period. A gradual decline in juvenile arrests throughout the period began before the police started conducting diversionary conferences. The time series for the cases disposed of informally
dur-ing the study showed no disruption from the pattern prior to the experiment, and this is the pool of offenders who would have been affected by net-widening. Because offenders were selected for this study after they had already been arrested, there was no discretion on the part of officers to determine which cases would be referred. Thus, there really was no opportunity for net-widening.
Dispositions of offenders three years prior to the study were compared to those handled by court during the period of the study and there were no important differences evident. Offenders di-verted tended to be less serious cases and were likely to have entered a guilty plea if the case had gone to court, thus slightly increasing the average seriousness of the cases remaining in the system. Overall case processing of juvenile offenders by police and the courts was largely unaffected by the existence of the program.
6. How does conferencing compare to the existing system and to other restorative justice
practices?
Existing System
Victims, offenders and offenders’ parents who participated in a conference were at least as satisfied with the way their case was handled and to have experienced fairness as those whose cases were processed through court (Exhibit S1). Victims and parents were more likely to feel that their opinion had been adequately considered. There were no significant differences between the control and treatment (decline and conference combined) victims for the satisfaction, fairness, accountability and opinion items. Among property crime victims, there was a
signifi-cant difference: the treatment group was more likely to say the of-fender was adequately held accountable for the offense.
Sixty-three percent of conferenced offenders said they were very satisfied with the way their case was handled, compared to 34% of the control group and 24% of the decline group. Similarly, parents were also more likely to say they were very satisfied with the confer-ence compared to the control or decline group parents. Parents of conferenced youth were more likely to report fairness in their child’s case than those disposed by courts. Still, a majority of all parents in the survey experienced fairness with the handling of their child’s case. Conferenced parents were more likely to have felt their
opin-ion had been adequately considered in their child’s case than parents of court-disposed offenders: 92% of the conference group, 84% of the control group, and 55% of the decline group.
Police-based conferencing produced outcomes for offenders more specifically tailored to the individual’s circumstances than the court process, especially for violent cases with personal victims. Outcomes from conferences were more likely to include community service as a reparative response and less likely to require monetary payments than outcomes from the courts.
Offenders Victims 93% 80% 97% control decline CONFERENCE 86% 96% 97% decline control CONFERENCE 73% 79% 96% decline control CONFERENCE Parents
Exhibit S1. Satisfaction with handling of case
Restorative Justice Programs
Other than conferencing, the primary restorative justice program for which there is research is victim-offender mediation (VOM). Participant questionnaires for the present study were designed from those asked in a majority of the VOM evaluations. There are a number of differences between confer-encing and mediation, though both utilize a voluntary collaborative model with the purpose of repair-ing the harm caused by the crime. Individual VOM programs also vary regardrepair-ing the type of cases qualifying and the source of administration and case referral sources.
Police-based conferences in the present study produced participant results and program participation rates higher than any of the reported VOM programs (Exhibit S2). The agreement compliance rates in the Bethlehem study are comparable to those cited in other mediation and conferencing studies. In light of these findings, concerns raised by VOM advocates that victims and offenders would be less trusting of police than of impartial community volunteers seem unfounded.
Finally, crude cost comparisons suggest that police-based conferencing is no more expensive than any of the VOM programs compared. Because police conduct conferences as part of their routine community policing activities, there were no additional program costs to the department beyond initial training costs.
General Conclusions
In summary, the following general conclusions can be made:
• Typical American police officers are capable of conducting conferences consistent with due process and restorative justice principles, given adequate training and supervision.
• While conferencing did not transform police attitudes, organization culture or role perceptions, it did move those with the most exposure to conferencing toward a more community-oriented, problem-solving stance.
• Police-facilitated restorative conferences can produce conflict-reducing outcomes, most clearly in cases of interpersonal violence. Because of a strong self-selection bias, this study could not confirm a reduction in recidivism due to conferencing. Like other voluntary diversion programs, cooperative cases participated, uncooperative cases did not.
• Victims, offenders and parents who participated accepted this police-based restorative justice response, as indicated by high rates of satisfaction with the process and experiences of fairness. • Police-facilitated restorative conferences produced participant satisfaction and perceptions of
fairness at least as high as other restorative justice programs and the courts. Participation rates and compliance rates for conferences were also comparable to other restorative justice programs.
29% 35% 40% 28% 42% Austin Oakland Minneapolis Albuquerque Bethlehem conferencing V O M s i t e s
Exhibit S2. Particiation rate comparison with VOM
1
Restorative Policing
This is a report on the Bethlehem Pennsylvania Police Family Group Conferenc-ing Project, National Institute of Justice research grant.* Juvenile offenders who quali-fied for the study were randomly assigned either to formal adjudication or to a diversion-ary “restorative policing” process called family group conferencing. Police-based family group conferencing employs trained police officers to facilitate a meeting attended by juvenile offenders, their victims, and their respective family and friends, to discuss the harm caused by the offender’s actions and to develop a plan to repair the harm. Victim and offender participation is voluntary. The effect of the program was measured through surveys of victims, offenders, offender’s parents and police officers and by examining outcomes of conferences and formal adjudication.
Family group conferencing (also called community conferencing) originated in New Zealand in 1989 under the auspices of the social welfare department as a means of divert-ing young offenders from formal adjudication. Conferencdivert-ing was substantially revised and pioneered as a community policing technique in Wagga Wagga, New South Wales, Australia in 1991 (Moore and McDonald, 1995). The “Wagga model” was introduced to North America in 1995 by the REAL JUSTICE® organization, and more than 2,000 po-lice, probation officers, educators and others in the United States and Canada have been trained as conference facilitators (Umbreit and Zehr, 1996; Wachtel, 1995).
Police-based conferencing provides a forum for the police to bring together juvenile offenders and their victims with their respective families and supporters. This micro-community of citizens directly affected by the crime collectively seeks resolution of the injuries, which may include apology, reparation to the victim, and reintegration of the offender. Ideally, solutions are not imposed by the facilitator, but instead result from the dynamic interaction of participants. Goals of the conference are: to encourage young of-fenders to achieve empathy toward their victims and take responsibility for their crimes, allow victims to move toward forgiveness and healing, and empower citizens to appropri-ately address their own local problems (McCold, 1997; Moore and O’Connell, 1994).
The practice of restorative policing is related to three trends in re-examining the Western system of justice: 1) community policing and problem-oriented policing (Goldstein, 1990); 2) reintegrative shaming theory (Braithwaite, 1989); and 3) restorative justice (Zehr, 1990; McCold, 1997a).
*This project was supported under award number 95-IJ-CX-0042 from the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice. Points of view in this document are those of the authors and do not necessarily represent the official position of the U.S. Department of Justice.
Community and Problem-Oriented Policing
Conferencing is more consistent with Herman Goldstein’s definition of problem-oriented policing than with the vague and varied notions of community policing often set forth by practitioners and researchers. Community policing means many different things to different people (Hunter and Barker, 1993; Bayley, 1994). There is some consensus that its general aims are to prevent crime and create a better quality of life and to change the reactive, control-oriented style of policing to a proactive, problem-solving, service-oriented style. The stated objectives of community policing, however, are varied: to re-duce crime, fear of crime, calls for service and complaints against police; to increase pre-ventative knowledge, crime clearance rates, public satisfaction, number of volunteers, police satisfaction, efficiency and effectiveness; and to build police-community partner-ships (Normandeau, 1993). The specific methods of achieving these aims may differ very little from previous policing approaches, and despite its popularity, community policing is described as “more rhetoric than reality” (Mastrofski, 1988; Klockars, 1988; Jones et al., 1994; Bull & Stratta, 1994; Stenson, 1993).
Herman Goldstein, considered by many to be “the father of problem-oriented polic-ing,” differentiates between community policing and problem-oriented policing. He says that community policing is “designed to place great emphasis on one great need in polic-ing, which is to engage the community” (1997, p.8); problem-oriented policing has a broader focus—to adopt an analytical approach to identify and solve the specific problems that police confront. A key element in this is intensively engaging the community in problem solving (Goldstein, 1990; 1997).
Goldstein argues that the job of social control in society ultimately depends upon net-works other than the police, netnet-works that the police can only facilitate and support. The community should become responsible for policing itself:
Several arguments can be made for maximum use of informal controls that are already available in the community. First, invoking informal nongovernmental con-trol may truly be the most effective means for dealing with the problem. Second, doing so reinforces the concept of the police as facilitators in getting the community to control itself rather than depending on the police and the criminal justice system for control. Third, it supports the strong preference, when an option exists, for us-ing the least restrictive, least intrusive method of dealus-ing with a problem. (Goldstein, 1990, p.121).
Goldstein offers numerous examples of police mobilizing the community and making use of existing forms of informal social controls, such as: involving citizens in developing solutions to specific crime problems; promoting interaction among populations of varying age and racial composition to reduce fear; holding meetings to resolve ongoing conflicts among neighborhood residents; and seeking the help of “those who, because they have some power over an individual, may be able to influence his or her behavior” (1990, p.121).
Goldstein claims that individual police tend to have a clear definition of community: In what I have observed of the practice, as distinct from the rhetoric of community policing, police tend to engage the citizenry in a very pragmatic and more relaxed manner. They use “community” rather deftly to describe those affected in any way by the specific problem they are attempting to address, or the program being launched in response to the problem. (1990, p.25)
It is the involvement of the micro-community of those affected by a specific crime in providing informal social control and developing a mutually acceptable plan for resolu-tion that makes family group conferencing consistent with Goldstein’s view of problem-oriented policing.
Reintegrative Shaming Theory
John Braithwaite’s (1989) theory of reintegrative shaming contributed to the de-velopment of the Wagga model of conferencing. This theory about the causes of crime in societies has two parts. The first part suggests that the manner in which a society handles the emotion of shame will determine its degree of crime and violence. When shame is used to humiliate or stigmatize, those who are stigmatized will seek out criminal subcul-tures where they can find positive self-images. According to the theory, there is a positive relationship in societies between the intensity of stigmatizing shaming and the preva-lence of crime, viopreva-lence and criminal subcultures.
The second part of the theory seeks to explain why people generally adhere to behavioral norms, turning the traditional “What causes crime?” question upside down. Braithwaite asserts that societies that use “reintegrative shaming” have lower levels of crime and violence. Reintegrative shaming involves encouraging wrongdoers to experi-ence shame for their actions while allowing them to maintain their dignity. This is accom-plished by holding wrongdoers accountable for their actions and providing them with an opportunity to make things right. Conferencing is designed to facilitate a process of rein-tegrative shaming.
Restorative Justice
Although developed independently from the restorative justice movement, confer-encing is considered an important new development in restorative justice practice. Re-storative justice views crime, not primarily as a violation of law, but as an offense against people and relationships. Restorative justice identifies three main stakeholders in crime: victim, offender and community. According to the philosophy, the community (however vaguely defined) has a responsibility in facilitating a restorative response to wrongdoing; that response should include holding offenders accountable for their actions and requir-ing them to make reparation to the victim and the community. Punishment and “just deserts” are not goals of restorative justice, and are viewed as ineffective, undesirable and counterproductive responses to crime (McCold, 1995).
Victim-offender mediation, restitution and community service programs have tra-ditionally been central practices for restorative justice advocates. With the advent of fam-ily group conferencing and sentencing circles, the restorative justice movement has rec-ognized the importance of including the personal communities of care of both offenders and victims in the resolution of criminal conflict (Umbreit & Zehr, 1996). Restorative justice practice is moving from excluding the micro-community under early victim-of-fender mediation models to including them as a central part of the restorative process (Van Ness and Strong, 1997; Wright, 1996).
Tony Marshall (1994) suggests that restorative justice seeks to reduce crime by strengthening bonds of interdependency while holding offenders accountable. Marshall defines restorative justice as:
a process whereby all the parties with a stake in a particular offence come together to resolve collectively how to deal with the aftermath of the offence and its implica-tions for the future. Parties with a stake in an offence include, of course, the victim and the offender, but they also include the families of each, and any other members of their respective communities who may be affected, or who may be able to contrib-ute to prevention of a reoccurrence. (Marshall in McCold, 1997b, p.2)
Restorative Policing
Police-based family group conferencing exemplifies a union of community and prob-lem-oriented policing (especially as conceived by Goldstein), reintegrative shaming and restorative justice—a union which could be termed “restorative policing” (McCold and Wachtel, 1998). As an operational philosophy for police, restorative policing seeks to:
1) Encourage accountability, reparation, reintegration and healing. 2) Reduce recidivism.
3) Resolve conflict and eliminate ongoing problems.
4) Provide communities with a satisfying experience of justice.
5) Reduce reliance on the criminal justice system and formal processes. 6) Transform police attitudes, organizational culture and role perceptions.
Some have expressed concerns and criticisms about police-based family group con-ferencing, which could be extended to restorative policing as a whole. Efforts to institute restorative policing programs such as conferencing should consider these as part of their evaluation. These concerns and criticisms include:
1) The focus on improving criminal justice responses distracts from the broader goal of tackling social injustice.
2) Conferencing may lead to net-widening.
3) Conferencing threatens principles of proportionality; that is, outcomes from con-ferencing may be too severe or too lenient compared to outcomes from formal justice processes.
4) Conferencing poses a risk of double jeopardy,
6) Police should not run conferences because they “have a coercive role, their legiti-macy is grounded in the invocation of punishment and they do not enjoy the respect of young people, especially young people from oppressed racial groups. So, it is argued, it is naive to believe the police could do a good job.” (Braithwaite, 1994, p.207).
Proponents of victim-offender mediation have articulated a number of possible dangers of police-based conferencing, similar to the concerns articulated above (Umbreit and Zehr, 1996):
1) Inadequate preparation could “significantly limit the impact of FGC in humaniz-ing the process in such a manner that parties feel safe and prepared to attend and participate freely in a genuine dialogue” (p.6).
2) Conferencing and conference facilitators may be insensitive to victims’ needs and coercive in encouraging their participation in the process.
3) Young offenders may be intimidated by adults and uniformed police officers; they may not feel safe or comfortable enough to share thoughts and feelings and to genuinely “own up” to the criminal behavior.
4) Police may be incapable of being neutral facilitators, falling into authoritarian behavior patterns and undermining the process of reintegrative shaming. 5) The scripted conferencing process may be too rigid and insensitive to cultural needs
and preferences within a community.
6) Police-based conferencing may lead to net-widening. Developing Hypotheses
The only completed empirical evaluation of restorative policing to date is a study of the program in Wagga Wagga, which used a before/after design (Moore, 1995). The “Wagga report” concluded that implementation of conferencing for juvenile offenders had decreased the number of cases being dealt with by formal processing in the court without increasing the overall recidivism rate. The introduction of FGC provided the police with an additional informal process beyond counsel and release, and changed the manner that police disposed of youthful offenders. The rate of referral to court was reduced from 51 percent to 28 percent following the introduction of conferencing. The results also sug-gested that the introduction of FGC was truly diversionary, without producing a net-widening effect.
The program in Wagga Wagga received widespread support from frontline police personnel and local community members (Graham, 1993;Moore, 1995, 1993; Moore and McDonald, 1995; Moore and O’Connell, 1994). The initial evaluation of the approach dem-onstrated that juveniles were able to be diverted from formal court processing without increasing the rate of recidivism. Crime victims found overwhelming satisfaction by be-ing actively involved in the process, and families were supported in their efforts to deal with the misbehavior of their children. Victim participation exceeded 90 percent, mutu-ally accepted restitution agreements were developed in 95 percent of conferenced cases,
and offenders complied with these agreements more than 95 percent of the time. The active involvement of the community in resolving juvenile crime altered both the view of police toward the community and young people, and the community’s view of the police (Moore, 1995).
Moore’s study had a number of inherent weaknesses. Due to the lack of a random-ized design, the group of offenders processed before the introduction of FGC were not strictly comparable to those processed after its introduction. Rates of re-apprehension were somewhat higher for those processed by the courts following introduction of FGC, and appeared to have remained unchanged for those processed informally by the police (warning versus conferencing). This suggests that re-offending was more a function of choice of processing than the effects of the conferencing, per se.
The few qualitative studies of the Wagga Wagga program have suggested that one of the most significant effects of conferencing was on the attitude that the police depart-ment had toward itself. These studies suggest that involvedepart-ment by the police in confer-encing produced a cultural shift from a punitive legalistic approach to a more problem-solving, restorative approach. Additionally, “. . .when police are involved with this more complex model [conferencing], they find it far more satisfying than the traditional alter-native” (Moore, 1995, p.212).
John Braithwaite, Lawrence Sherman and Heather Strang are currently collabo-rating in the Reintegrative Shaming Experiment (RISE) in Canberra, Australia (Sherman, 1996; Sherman and Barnes, 1997; Sherman and Strang, 1997a, 1997b; Strang, 1997; Strang, H. & Sherman, 1997). The RISE project is randomly assigning juvenile offenders and adult “drink driving” offenders to police-run “community accountability conferences” or to traditional court. They are conducting in-depth evaluations of participants’ percep-tions, victim and offender background information, and systematically observing both the conferences and the court processes. The results of RISE will be an important supple-ment to the present study and should allow for a cross-national comparison of police-based conferencing.
The primary purpose of the present study is to evaluate the implementation of conferencing as a restorative policing practice, examining the effects of the practice on police and the community and comparing those results to equivalent data on formal adju-dication and other restorative justice approaches. Reflecting the goals of restorative po-licing and the concerns about police-based conferencing previously described, the present study asks the following questions:
1. Can typical American police officers conduct conferences consistent with due pro-cess and restorative justice principles?
2. Does conferencing transform police attitudes, organizational culture and role per-ceptions?
3. Does conferencing produce conflict-reducing outcomes by helping to solve ongoing problems and reduce recidivism?
4. Will victims, offenders and the community accept a police-based restorative justice response? 5. How does the introduction of conferencing alter the case processing of juvenile
offenders?
6. How does conferencing compare to the existing system and to other restorative justice programs?
Chapter 2 describes the Bethlehem Police Family Group Conferencing Project in detail. Chapters 3 through 8 examine the present research in light of these six questions. The final chapter draws together the conclusions from the other chapters to address what can be known about these questions from the present findings.
2
The Bethlehem Experiment
In the summer of 1995, the Bethlehem Police Department and the Community Service Foundation (a private not-for-profit organization) began planning a two-year re-search partnership to study the effectiveness of police-based family group conferencing. Sponsored by the National Institute of Justice*, the study began November 1, 1995, after a three-day REAL JUSTICE® training for 18 Bethlehem Police officers, conducted by three Australian pioneers in family group conferencing.
The Bethlehem Police Department has 140 sworn police officers and is actively involved in addressing the needs of city residents. The department has ongoing crime prevention and community policing programs which include four permanent substations, a mobile substation, bicycle patrols, and four full-time officers assigned to middle schools. The city of Bethlehem is located in southeastern Pennsylvania, a two-hour drive west of New York City and a one-and-a-half hour drive north of Philadelphia. Bethlehem, Allentown and Easton comprise a three-city metropolitan area, surrounded by approxi-mately 25 townships and boroughs of varying sizes. Bethlehem has an area slightly over 19 square miles with a population of approximately 72,000. It is part of both Northampton and Lehigh counties.
Before the experiment began, the department began a vigorous marketing effort to gain the community’s support for the diversion program, which included presentations to service organizations, merchant associations, school administrators and church groups. Several articles appeared in the local newspapers.
Over the course of the experiment, the 18 police officers participating in the pro-gram had quarterly meetings to review the progress of the propro-gram, identify and resolve problems and be appraised of current research statistics. The group operated as a self-directed work team with a senior officer as liaison (program liaison officer) between the department, courts, probation and schools. The group formulated a program name, “Op-eration P.R.O.J.E.C.T.” (Program for Redirection of Offending Juveniles through Empa-thy-building and Conferencing Techniques), and developed a mission statement and goals for the program. The mission statement reads:
The Bethlehem Police Department’s “Operation P.R.O.J.E.C.T.” is an alternative justice program for juvenile offenders and their victims. By providing a forum for victims to express feeling and take part in the repair of harm, the offenders must own and evaluate their behavior and how it affects other people.
The program goals the officers articulated were satisfying victims, repairing harm/dam-age, re-educating juvenile offenders, offenders “owning” their behavior, lowering recidi-*This project was supported under award number 95-IJ-CX-0042 from the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice. Points of view in this document are those of the authors and do not necessarily represent the official position of the U.S. Department of Justice.
vism rates, providing an alternative to punishment, increasing community satisfaction, and reducing court system workload.
The following section describes the methods used for selecting cases for the study and soliciting participation in the program. A presentation and discussion of case selec-tion results and participaselec-tion rates follows.
Methods
The Bethlehem Police Department adopted the following policy for eligibility in the juvenile diversion program:
• Only juveniles arrested by the Bethlehem Police Department are eligible.* • Only first-time offenders are eligible. (For the purpose of the research project, a
first-time offender is defined as a juvenile who has not been through the juvenile probation system.)
• No felony level crimes are eligible unless specifically agreed to by the chief of juvenile probation.
• No drug or alcohol crimes (possession or delivery) are eligible. • No sex offenses are eligible.
• Only assaults which meet the following conditions are eligible: a) graded as simple assaults (or threat or harassment) where:
1) there is no serious bodily injury 2) no weapons were used
3) juvenile assaults a juvenile where there is less than a 5- year age gap b) graded as a summary violation
• Thefts of a misdemeanor or summary level are eligible.
• Property crimes of a misdemeanor or summary level are eligible.
The above policy and guidelines were established after conferring with the juve-nile court judges, district attorneys, and chief juvejuve-nile probation officers from both Lehigh and Northampton counties. Because this was a pilot program, the cases selected for in-clusion were confined to the least serious cases available. A limited number of prior sum-mary arrests did not automatically disqualify a juvenile from the program, allowing for some discretion by the program liaison officer, in consultation with the principal investi-gator. Cases where the only identifiable victim was the arresting officer—so-called “con-tempt of cop” cases—were not included in the study, at the request of the principal inves-tigator.
The program liaison officer regularly reviewed arrest records submitted by officers over the course of the experiment—November 1, 1995 through May 1, 1997—earmarking cases that appeared to qualify for the study. Criminal history information was then checked to confirm eligibility. The liaison officer then phoned the principal investigator to submit the selected cases to random assignment.
There were to be a total of 150 property cases and 75
crime-against-* As of October 1996, this requirement was amended with the Hellertown Police Department joining the project.
Hellertown is a smaller jurisdiction adjacent to Bethlehem. The Hellertown Police contributed one case to the experiment.
person cases selected for the experiment, two-thirds assigned to a treatment group and one-third assigned to a control group. Cases were defined as a criminal incident, and each case could involve multiple offenders.
Two lists of random numbers had been generated (n = 150 and n = 75), each with equal distributions of integers 1, 2 and 3. The cases were placed on a list in the order they were reported to the researcher. All incidents with 1’s were coded as the control group, 2’s and 3’s the treatment group. The random assignment list was never revealed to the liai-son officer, and he could not anticipate the next assignment. When a case was included, a determination was made whether the offense was primarily a crime-against-property or primarily a crime-against-person, hereafter referred to as property crimes and violent crimes. The case information was then entered into the next line of the relevant random list and the liaison officer was informed whether the case was control or treatment.
None of the victims or offenders for control group cases were informed about the existence of the diversion program. These cases were left to be processed without police diversion. When a case was assigned to the treatment group, the liaison officer began to attempt to contact the offender(s) involved, explain the program and elicit their partici-pation. Because the program is mindful of the due process rights of offenders, they must understand that they can opt out of the program and decide to face court with all their rights intact. Acceptance into the program required the offender to admit non-inculpa-tory responsibility for the charge (actus reus). When offenders agreed to participate, the liaison officer then contacted the victims to explain the program and elicit their partici-pation. Only where both offender and victim were willing to participate was the case assigned to the facilitating officer. If either party was unwilling to participate, the case was not conferenced and, thus, was processed through normal channels like the control cases. After the key parties agreed to participate, the facilitating officer then had the responsibility of inviting other supporters, arranging a time and place to meet, and speak-ing with all participants to prepare them for the conference.
Results
There were 1,285 juveniles arrests between November 1, 1995, and May 1, 1997 (excluding traffic violations). As reflected in the selection criteria above, the Bethlehem Police Department wanted to exclude repeat offenders, offenders charged with felony offenses, and those not residing locally; the county prosecutors wanted to exclude drug and alcohol offenses, sex offenses, weapon offenses, and assaults where offenders used a weapon or were more than four years apart in age from the victim; the principal investi-gator wanted to avoid crimes where the only available victim was the arresting officer and cases disposed without formal arrest. These are not mutually exclusive categories. For example, 56 percent of the offenders charged with a felony were also disqualified
because of prior records.
As shown in Exhibit 1, prior referral to juve-nile probation (i.e., prior misdemeanor or felony ar-rest) was the largest single reason for disqualifying cases from the study, with 29 percent who could have been disqualified for that reason alone. Overall, 35 percent of offenders could have been disqualified for one or more of the three prior history reasons (prior referral to juvenile court, more than 3 prior arrests or adjudicated youth), 18 percent for inappropriate crimes (drug and alcohol or public order offenses), 15 percent for non-Bethlehem-residency, 14 percent for too trivial an offense (handled informally), and 10 percent for too serious a crime (felony, weapons or sexual offenses).
There were 227 juvenile arrests during the study—18 percent of the total number of arrests (n = 1,285)—who could not be disqualified based on any of the known reasons stated above. Some of these cases were not selected because of offenders who were charged with a simple assault in spite of using a weapon or who had committed disorderly conduct without a victim other than the arresting officer. This de-tailed information was only available from the arrest reports, and thus these offenders could not be disqualified based upon information available from the computerized records. Nonetheless, this apparently qualifying but not selected group of offenders will provide a useful comparison group later in this
re-port.
In order to compare the proportions of cases disqualified for different reasons, it is necessary to create mutually exclusive categories. If cases are first disqualified cause the crime was too serious, then be-cause the offender’s prior history was too serious, then because the offenses were in-appropriate, then because the case was handled informally, and finally because the offender was not a local resident, an ap-proximation of proportions of cases dis-qualified by reason can be considered.
non-selected eligible crime too serious offender history crime inappropriate handled informally non-resident selected Exhibit 2
Offender eligibility categories n = 1,285 18% 23% 4% 7% 11% 28% 9% Exhibit 1
Proportion of total juvenile arrests disqualified from study
7% 2% 1% 29% 15% 11% 13% 5% 14% 15%
felony offensesweapons chargessexual assaultsprior referral to juvenile courtmore than 3 prior arrestsadjudicated youthdrug/alcohol offensespublic order offenseshandled informallynonresident by reason for disqualification
As shown in Exhibit 2, offender prior history remains the most common disquali-fier (28 percent), followed by inappropriate offenses (11 percent) and crime seriousness (9 percent). Eighteen percent of all juvenile arrests were disqualified for reasons not avail-able in computerized records. Thus, during the course of the experiment, 215 criminal incidents involving the arrests of 292 juveniles qualified for the study, representing 23 percent of all juvenile offenders arrested by the Bethlehem Police during the time period. Qualifying cases were submitted to random assignment until the target rate of 75 violent crimes was achieved. The project included 140 property crimes, 93 percent of the 150-case target. The results of the random assignment are shown in Exhibit 3. The ran-domized experimental assignment was adhered to in all 215 cases in the study, achieving the one-third/two-thirds assignment with less
than .24 percent deviation. The number of offend-ers per case ranged from 1 to 6, with an overall average of 1.36 persons-per-crime (ppc). This ra-tio did not distribute equally across the
experi-mental groups. Among violent cases, treatment and control groups were about equally as likely to involve multiple offenders with 1.40 ppc in the control group and 1.52 ppc in the treatment group. However, among property cases, the treatment group, with 1.22 ppc, was less likely to involve multiple offenders than
the control group, with 1.45 ppc. Thus, treatment group property cases were less likely to have in-volved multiple offenders than cases in the other groups. As a result, the assignment of offenders
slightly deviates from the case distribution as shown in Exhibit 4, though not by a statis-tically significant amount.
The participation rates in the program for those cases assigned to be conferenced (treatment group) varied, with 32 percent of violent cases and 52 percent of property cases participating. Taking into account that there were multiple offenders for some cases, the offender-based participation rates were slightly lower, with 32 percent of violent offenders and 50 percent of prop-erty offenders participating, as shown in Exhibit 5. Thus, two-thirds of the violent offenders and half of the property offenders selected for experimental treatment were actu-ally processed the same as those selected for the control group, through the traditional court processes, χ2 (1, n =
189) = 6.06, p < .05.
It was expected that the randomly assigned groups
Treatment group participation rates Exhibit 5
crime type
violent property total
50% 32% 42% Exhibit 3 violent property column
Random assignment results - Cases control treatment 25 47 72 33% 34% 33% 50 93 143 67% 66% 67% total 75 140 215 100% 100% 100% Exhibit 4
Random assignment results - Offenders violent property column control treatment 35 68 103 32% 38% 35% 76 113 189 68% 62% 65% total 111 181 292 100% 100% 100%
would be similar in all important respects. Because a proportion of those cases selected for treatment failed to result in a conference, the assumption of equality of groups is in doubt. Thus, there are essentially three experimental groups—the control group, the con-ference group (treatment/participation) and the decline group (treatment/non-participa-tion). Knowledge of any ways the groups differ will be helpful in interpreting results from this study.
The specific criminal offenses included in the sample are shown in Exhibit 6. Among property crimes, 74 percent were retail theft cases. Among the violent crimes, 88 percent were harassment or disorderly conduct. Harassment-by-communication is included in the harassment category, and all other crimes are collapsed to produce the specific crime groupings used in this study. Thus, property crimes are divided into retail theft and other property crimes, and violent offenders are divided into three crime types—disorderly, harassment and other violent crimes—as shown in Exhibit 7.
Among property crimes, retail theft cases were somewhat under-represented in the control group and over-represented in the conference group. Participation rates were
Exhibit 6
Arrest charges for offenders included in study by experimental group
total control conference decline retail theft 129 44% 44 43% 48 60% 37 34% criminal mischief 24 8% 7 7% 6 8% 11 10% theft by unlawful taking 10 3% 4 4% 1 1% 5 5% receiving stolen property 4 1% 1 1% 3 3%
criminal trespass 4 1% 4 4% disorderly conduct 4 1% 4 4%
park after hours 2 1% 2 2%
theft/failure to deposit 1 0% 1 1% institutional vandalism 1 0% 1 1% false alarm 1 0% 1 1% unauth.use of vehicle 1 0% 1 1% Property Subtotal 181 62% 68 66% 56 70% 57 52% disorderly conduct 49 18% 18 21% 10 13% 21 19% harassment 38 13% 12 12% 6 8% 20 18% simple assault 20 7% 2 2% 7 9% 11 10% terroristic threats 2 1% 2 2% harass.by communication 1 0% 1 1% noise a nuisance 1 0% 1 1% Violent Subtotal 111 38% 35 34% 24 30% 52 48% Total 292 100% 103 100% 80 100% 109 100% Exhibit 7
Arrest charge categories for offenders included in study by experimental group
total control conference decline
retail theft 129 71% 44 65% 48 86% 37 65% other theft 52 29% 24 35% 8 14% 20 35% Property Subtotal 181 100% 68 100% 56 100% 57 100% disorderly conduct 49 18% 18 21% 10 13% 21 19% harassment 38 13% 12 12% 6 8% 20 18% other violent 25 23% 6 17% 8 33% 11 21% Violent Subtotal 111 100% 35 39% 24 30% 52 48% Total 292 100% 103 100% 80 100% 109 100%
58 percent for retail theft and 29 percent for other property crimes.Among violent offend-ers, the crime subcategories are more equally distributed across all three experimental groups, and the participation rates were comparable, with 35 percent for disorderly, 32 percent for harassment, and 40 percent for other violent crimes. Thus, the only difference in the distribution of crime subcategories across experimental groups was that among property offenders those offenders attending a conference were more likely to be charged with retail theft versus other
property crimes than those of-fenders in the control or de-cline groups.
Exhibit 8 shows the ex-perimental groups broken down by age, seriousness of charge, race/ethnicity, gender and residence. Differences be-tween experimental groups in seriousness of charge and race/ ethnicity are not statistically significant controlling for crime type. However, differ-ences in age, gender and resi-dence are statistically signifi-cant after controlling for crime type. Thus, the experimental groups differed in several re-spects. Among violent offend-ers, males comprised half the conference group and 75 per-cent of the decline group. Among property offenders, the decline group had a lower pro-portion of 13-year-olds and a higher proportion of Zip1 resi-dents than the conference and decline groups; the property control group had a higher proportion of Zip3 residents than the property decline and
Experimental Group Comparisons
under13 number age13 age14-15 age16-17 summary misdemr-3 misdemr-2 felony-2 white black Latino other χ2 ,p Exhibit 8 male female one two >two none one two >two Zip1 Zip2 Zip3 Other
Most serious current arrest Age on arrest
Race/Ethnicity
Total Violent Property
24% 109 16% 32% 28% 93% 0% 7% 0% 35% 14% 50% 2% 69% 31% 82% 14% 5% 75% 17% 6% 1% 48% 34% 13% 6% 19% 52 23% 25% 33% 87% 0% 13% 0% 31% 13% 54% 2% 75% 25% 81% 10% 10% 65% 25% 8% 2% 50% 31% 13% 6% 28% 57 9% 39% 25% 98% 0% 2% 0% 39% 14% 46% 2% 63% 37% 82% 18% 0% 84% 11% 5% 0% 46% 37% 12% 5% Number of current charges
Number of prior arrests
Residence of offender 25% 103 29% 34% 12% 13.2 89% 5% 3% 2% 18.3 44% 6% 49% 2% 15.0 71% 29% 7.8 88% 9% 3% 78% 17% 6% 0% 28% 33% 36% 3% 24.1 6.3 6.3 23% 80 29% 31% 18% 0.04 98% 0% 3% 0% 0.02 41% 1% 51% 6% 0.02 53% 48% 0.02 94% 5% 1% 83% 15% 0% 3% 33% 36% 20% 11% 0.00 ns ns 31% 68 29% 28% 12% 11.4 91% 6% 1% 0% 47% 6% 44% 3% 65% 35% 87% 9% 4% 82% 13% 4% 0% 26% 31% 40% 3% 18.2 8.6 7.9 1.8 7.4 7.4 23% 56 25% 36% 16% 0.08 98% 0% 2% 0% 46% 2% 46% 5% 54% 46% 93% 5% 2% 91% 9% 0% 0% 30% 36% 21% 13% 0.01 ns ns ns ns ns 14% 35 29% 46% 11% 86% 3% 6% 6% 37% 6% 57% 0% 83% 17% 8.1 91% 9% 0% 69% 23% 9% 0% 31% 37% 29% 3% 9.7 8.9 8.5 6.8 6.8 5.6 21% 24 38% 21% 21% 96% 0% 4% 0% 29% 0% 63% 8%
decline decline decline control conference control conference control conference
50% 50% 0.02 96% 4% 0% 63% 29% 0% 8% 38% 38% 17% 8% Gender ns ns ns ns ns ns χ2,p χ2 ,p χ2 ,p χ2,p χ2,p χ2 ,p
conference groups. There was no discernable differences in the three Zip codes which might explain these differences in participation by locality (residents differed as much within zip code as between zip codes in S.E.S. and racial composition). In all other re-spects, based on the available data, the decline group is statistically similar to the confer-ence group, and the control group is statistically similar to the conferconfer-ence and decline groups.
Understanding the reasons for declining to participate may help ex-pose a self-selection bias should one exist. Because conferencing is volun-tary, either victim or offender could de-cline to participate. Other factors be-sides direct refusals to participate could also result in failure of confer-ences to proceed. Cases were coded by the reason for declining to participate
as shown in Exhibit 9a. Among the 52 selected violent offenders not conferenced, offend-ers declined in 29 percent of the cases and the victim declined in 62 percent of the cases. Among the 57 selected property offenders not conferenced, the corresponding decline pro-portions were 70 percent for offenders, and 14 percent for victims. Thus, offenders were much more likely to be the reason for declining conferencing in property offenses, victims more likely in violent offenses.
The program participation rate is the number of conferences divided by the num-ber of cases selected in the treatment group. In order to calculate individual victim and offender participation rates, it is necesary to limit the categories in the denominator of the rate. For example, offenders who could not be found could not have agreed or declined to participate, so they should not be counted in the offender decline figure. Likewise, victims of offenders who declined to participate were never ask to participate, so the number of cases
for victims to available to de-cline is limited to the number of con-ferences (where of-fender has agreed to participate) plus the number of victims declining.
violent property offender declined
contests charges prefers court
reoffend prior to contact unable to contact victim declined victim declined victim nonresponsive case excluded 29% 62% 10% 70% 14% 16% settled prior to contact
administrative error column 15 5 6 1 3 32 14 18 5 4 1 52 40 7 14 5 14 8 8 0 9 5 4 57 total 50% 37% 13% 55 12 20 6 17 40 22 18 14 9 5 109
Reasons for cases declining to participate
Exhibit 9a
Corrected participation rates
total violent property formulas total selected for treatment 189 76 113
total cases declined 109 52 57 excluded cases 14 5 9 reoffend prior to contact 6 1 5 offenders unable to contact 17 3 14 offender decline 32 11 21 victim decline 40 32 8 conferenced 80 24 56
offender participate rate 79% 84% 75% =(vic+conf)/(off+vic+conf) victim participate rate 67% 43% 88% =(conf)/(vic+conf)
total participation rate 53% 36% 66% =conf/(off+vic+conf) program participation rate 42% 32% 50% =conf/total selected
